Voltage to current amplifier

[Domitronic]

Hello,

i would like to build a voltage to current amplier. What approach would you recommend for these parameters:

Input: AC voltage with +/- 2V with frequency range 30Hz to 3kHz

Output: AC current with +/- 150mA into grounded load resistor between 0.5 and 8 Ohms.

Power supply is +/-5V. The ideas i'm thinking of are

- Use an Opamp with high output current. Limit the current with a resistor at the output of the opamp to +/-150mA. Problem is that opamps with high current output seem to be rare.
- Use some kind of audio amplifier IC.
- Use a standard opamp with transistor amplifier afterwards. For example an AB amplifier with 2 transistors.

What would you recommend and what other options are available?

Thanks for your help!
 

[w2aew]

Does the amplifier output indeed need to be a current source - i.e. high output impedance, and deliver a specified current into to the 0.5 to 8 ohm load?

or

Do you really just need an amplifier capable of delivering a specified VOLTAGE into the 0.5 to 8 ohm load (thus supplying a high current)?

Your query sounds like you're asking for a voltage to current source solution (first option), but your suggested solutions sound like the 2nd option.

[magic]

- Use an Opamp with high output current. Limit the current with a resistor at the output of the opamp to +/-150mA. Problem is that opamps with high current output seem to be rare.

If that sort of solution is adequate then look for some NJM4556 - an old dual high current opamp for headphones and stuff like that. They can be found fairly cheap at stores for hobbyists, audio, AV gear repair or on evilBay (prefer local sellers, not China). Or at Mouser.

Connect both channels in parallel with separate resistors to the load, it should be able to do just a bit over 150mA that way. Or buy two of them, four channels will definitely suffice, even for 150mA RMS. (not sure if it's RMS or peak?)

If you need accurate current regulation, there is a trick for that too involving a current sense resistor between the amp and the load and a four resistor differential feedback network comparing its voltage drop against the input signal.

[Vovk_Z]

A Howland current pump is a good choice for AC voltage-to-current convertor and a grounded load.
150 mA is not too much for some powerfull opamps like LM6172 or similar (e.g. xDSL drivers) AD8397.
+-5 V power supply is too low for audio amplifier IC.

[magic]

NJM4556 will run just fine on ±5V supplies with ±3V output swing.

It's also cheaper and less prone to oscillation (very stable, in fact, even faced with capacitive load) than high speed opamps.

[Domitronic]

Actually a defined current would be best. And short circuit protection would be a plus as well. So i would prefer a solution with a current source / high output impedance if this is possible with limited effort.

Thanks for your replies. I will look up the suggestions.

[David Hess]

I agree with Vovk_Z; a Howland current pump will be ideal but a high output current operational amplifier or operational amplifier with a current booster will be necessary.

[Vovk_Z]

A high output current opamp can be googled by TS by googling "high output current opamp".
For example, I found cheap AD8531/8532 with 250 mA output current. But it is made for low voltage power supply - 6VDC max (+- 2.5-3.0 VDC), and has quite high input bias voltage.

[magic]

Using an opamp with less than 200mA output capability ticks the "overcurrent protection" box

Two channels of a dual can be paralleled by replacing the current carrying resistor with two resistors twice as high and inserting the other opamp as unity gain follower between the primary opamp's output and one of the resistors. This works for any topology.

[Terry Bites]

Here are some typical configurations https://www.google.co.uk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwjk2aaP04v2AhVSolwKHciLC7gQFnoECAQQAQ&url=https%3A%2F%2Fwww.ti.com%2Flit%2Fpdf%2Fslyy054&usg=AOvVaw1YP0QbspxjlfhkaAcL39PN


Here's work up of a practical circuit to try.

[Domitronic]

Hello,

based on the suggestions i tried to acquire some knowledge about the Howland current source. So i did a simulation in PSpice. The Opamps might not be the ideal choice but it is one that was in the libraries. Below is the schematic and the simulation result. Maybe you can help me by confirming or rectifying my observations:

- R3 defines the transfer ratio between input voltage and output current. So 2.5V / 0.15A equals 16.67 Ohms. I chose 18 Ohms as it is an E12 value.
- R4 defines the required output voltage swing of U2. What else do i need to consider when choosing a value?
- In the range of output resistor values i have most of the current is delivered by U1. With resistor values > 8 or 10 Ohms this changes and U2 is the one to deliver more current.
- The ratio R2/R1 needs to be the same as R4/R3. I have chosen 100 times higher values to avoid high currents. What else do i need to consider when choosing a values for R2/R1?

Any help is appreciated!

[langwadt]

A Howland current pump is a good choice for AC voltage-to-current convertor and a grounded load.
150 mA is not too much for some powerfull opamps like LM6172 or similar (e.g. xDSL drivers) AD8397.
+-5 V power supply is too low for audio amplifier IC.

opa547

[magic]

- In the range of output resistor values i have most of the current is delivered by U1. With resistor values > 8 or 10 Ohms this changes and U2 is the one to deliver more current.

I'm pretty sure the exact point where the change occurs is exactly half of R3 and R4, i.e. 9Ω.

I don't like this unequal load sharing. I was thinking about a circuit like below. Here R1 is a current sense resistor and U1 with its feedback networks is a differential amplifier, applying 1/2.7 of the input voltage across R1. U2 and R2 simply double U1 output current without U1 even needing to know about it. There is one subtle error source: a little current flows through R3,R4 right into the load, bypassing the measurement resistors. It could be ignored or accounted for.

edit
I think you could optimize the Howland pump for equal load sharing at less than 9Ω by changing the ratio of R1/R2 and R3/R4 to something other than unity. But it will always be exactly matched only at one particular load resistance.

- R4 defines the required output voltage swing of U2. What else do i need to consider when choosing a value?

I suppose: mind output current vs voltage capabilities of the opamp and how power dissipation is distributed between the opamp and the resistor. Power resistors are cheaper than power opamps, although in this particular case total power is still fairly low.

[Vovk_Z]

Any help is appreciated!

Look at my example. 2 k resistor value goes from 2 V input voltage, 150 R value goes from 150 mA output current. So we have to use 1 R shunt with those 2000/150 values. But we may use another values for Rshunt, approximately from 0.5 R to 8-10 R (load resistance is a base point). With too low Rshunt we'll have too low voltage across the shunt which is inconvenient. So, Rshunt have to be of a value which gives large enough shunt voltage, e.g. somewhere between 0.1-1.0 V (0.2-0.5 V possibly is better), but not too large.
With too high Rshunt resistance we need additional buffer for shunt voltage (as shown on my next post).
Output amplifier may consist of two or four opamps (as shown above), so one or two cases.
Example #1 is a basic circuit, and Example #2 is more close to real, with a trimmer to adjust output current and additional buffer for maximum possible output impedance. To imply trimmer we have to increase 150 R resistor values proportionally.

[Vovk_Z]

One more article for info: A Large Current Source with High Accuracy and Fast Settling
And one more How to Design a Precision Current Pump with Op-Amps

[Zero999]

Add pass transistors to an ordinary op-amp and use a Howland topology.


LTSpice file
https://www.eevblog.com/forum/projects/30a-precision-current-source/msg2324862/#msg2324862

Also see this thread:
https://www.eevblog.com/forum/projects/need-help-with-bi-directional-constant-current-source-(100ma)/

[Domitronic]

Thanks for all the feedback. Quite some reading and simulation to do now to get a good understandig of all the suggestions.

[Zero999]

Here are some typical configurations https://www.google.co.uk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwjk2aaP04v2AhVSolwKHciLC7gQFnoECAQQAQ&url=https%3A%2F%2Fwww.ti.com%2Flit%2Fpdf%2Fslyy054&usg=AOvVaw1YP0QbspxjlfhkaAcL39PN
(Attachment Link)

Here's work up of a practical circuit to try. (Attachment Link) (Attachment Link)

An instrumentation amplifier is a good idea. It can be improved a little by adding a buffer before the feedback resistor, although a compensation capacitor might be required.

[David Hess]

An instrumentation amplifier is a good idea.

The instrumentation amplifier and difference amplifier variations have the advantage of not requiring a set of matched resistors.

[Zero999]

Here's another way to boost the output current of an instrumentation/differential amplifier. U2 is a buffer with a high current capacity. It can be another op-amp, with a couple of pass transistors.


The output current is proportional to the input voltage and inversely proportional to R1 & R2 in parallel. The output current from U1 flows through R1 and from U2 through R2.
I_OUT = V_IN/(R1||R2)

[Terry Bites]

The big issue you face with all these curcuits is dynamic range and stability. All of them have the same core- a differential amplifier with a current sense resistor feeding back into the ref terminal. So they are all Howland sources in effect. Have a read of https://www.ti.com.cn/cn/lit/pdf/sboa437?keyMatch=SBOA437 to see the evolutionary paths.

As you push the ref terminal away from ground, the CM range of the opamp/ isntamp is reduced. On low voltage supplies this problem is further exacerbated. Reliable external current boosters cost compliance headroom and make it even harder to stabilise the loop. Instamps are by and large too slow to achieve a decent transient response.

But it depends on your needs. What kind of precision and step response is good enough?

[David Hess]

There are also configurations with a true transconductance (current) output so they preserve high output impedance at higher frequencies without the fall due to limited gain-bandwidth product.

[Terry Bites]

Sadly, I've yet to see an OTA/ conveyor with half way decent DC specs- well, outside a serious bucks sourcemeter.

For this thread I'm thinking some meaty little amps will do the job. Same ol' same ol' idea in a small space.